Researchers at Guangzhou’s Sun Yat-sen University Cancer Center have made a significant breakthrough with the development of an artificial intelligence (AI)-based model designed for the early detection of ovarian cancer. This cost-effective AI model aims to assist in diagnosing the deadliest form of gynecological cancer worldwide. In China, early-stage diagnosis rates for ovarian cancer are below 48%, with a 5-year survival rate of only around 40%.
As detailed in The Lancet Digital Health, the team developed a multi-criteria decision making-based classification fusion (MCF) risk prediction framework. The model was trained on data from over 6.7 million laboratory tests conducted at three Chinese hospitals over a period of nine years. It integrates data from 98 different routine laboratory tests, including blood, urine, biochemistry, and coagulation tests, which can potentially identify ovarian cancer cases by revealing biomarkers such as blood albumin concentration and lymphocyte ratio.
The MCF model was retrospectively tested on 10,992 women, with and without ovarian cancer, enrolled across the three hospitals between 2012 and 2021. The study authors report that the MCF model outperformed traditional ovarian cancer biomarkers like CA125 and HE4, as well as advanced models for ovarian cancer prediction, particularly in identifying early-stage cancers. The MCF also demonstrated acceptable prediction accuracy even when excluding high-ranking laboratory tests like CA125 and other tumor markers. The laboratory tests involved in the MCF are more cost-effective than the current diagnostic methods, which include routine tumor marker panel tests along with ultrasound, CT, or MRI. Full study details can be found in The Lancet’s article titled, “Artificial intelligence-based models enabling accurate diagnosis of ovarian cancer using laboratory tests in China: a multicentre, retrospective cohort study.”- Flcube.com
